No Sensation of Falling 38K Feet in 3.5 Minutes?

Factual Questions
61

There is a mode called direct law that effectively gives the pilots full authority. This isn’t fly-by-wire in the sense that fighters are, where they’re designed to be unstable and are essentially unflyable without computer assistance. The fly-by-wire in airliners is designed to provide protection from flight limits such as g loading, stalling etc, and to make flying easier under normal conditions. An airbus will auto-trim for you, so you use the sidestick to achieve the attitude you want, let go and the attitude will stay there. If you remove the fly-by-wire aspect it just turns into a normal aeroplane without protections, it doesn’t become in any way unflyable.

The problem with all simulators, from full motion Cat D sims used for airline training to PC based flight sims, is that they don’t have the data to be able to simulate conditions at the edges of the flight envelope. There is currently no way to accurately simulate what happened to AF447 other than to take a couple of test pilots up in a real aeroplane. That may seem criminal. Surely you should be able to accurately train pilots in recovery from a full stall. Well it would be nice if you could but the technology is not available and so instead they give you full stall training in small aircraft but go on to say DO NOT STALL! To help pilots with this they design the large aeroplanes to not let them stall. They put stick shakers in to alert of an impending stall and allow recovery before the stall occurs, and they put stick pushers in that physically push the control column forward which does two things, it prevents a full stall from occurring, and it provides the pilot with a simulation of the stalling characteristics of docile training aircraft. In Airbus’s case they have a computer that moderates the pilots inputs and if it sees anything it doesn’t like it says “nope, can’t do that sorry.” Having said all that, the pilots can ignore the stick shaker, they can fight the stick pusher (ala Colgan Air), and they can put an Airbus into Direct Law which turns it into a normal aeroplane with no protections. Also, if the aeroplane is faulty and multiple redundant systems fail, all bets are off. There are no good answers when all the backups have failed.

You would normally power out of a stall warning (i.e. when the stick shaker goes but the aeroplane hasn’t stalled yet), but thrust gives a nose up pitching moment and if you are having trouble getting the nose down to reduce the angle of attack one thing you’d want to do is get rid of the thrust.

At the start of the problems the aircraft went into Alternate Law and the auto thrust disconnected, Alt Law doesn’t provide any protections other than to pitch the nose up gently at high Mach numbers and to pitch the nose down gently at low airspeeds, both of those protections can be over-ridden by the pilots. In short, they should not have had any issue fighting the aeroplane.

The engines aren’t positioned in a way to put airflow over the control surfaces. Also the engines are most efficient at high RPMs and the airframe is most efficient at a particular moderate indicated airspeed. Indicated airspeed drops rapidly compared to true airspeed as altitude increases, for example an indicated speed of 250 knots is approximately a true airspeed of 380 knots at 30’000’. The general concept with cruising is to fly high where the thrust produced from a high engine RPM coincides with an efficient airspeed. There are margins though, you have margins above the stall and margins below speed of sound related effects. The margins may not be as large as what you might expect but they are by no means small. At all times the aircraft should be able to manoeuvre and stay a safe margin above the stall and below the mach buffet speeds. If turbulence is forecast then you fly lower where you have even more margin. When the pilots said, before the accident, that they couldn’t get to a higher level, it doesn’t necessarily mean they literally didn’t have the performance to get there, it may mean they can’t do it and maintain their safety margins. As the aircraft gets lighter the stalling speed drops and they can go higher, they also have more excess power available.

I think something that might be confusing to the layman is that nothing in flying is simple. The airspeed indicator doesn’t actually tell you your airspeed, the altimeter doesn’t tell you how high you are, the compass doesn’t tell you which way north is, power doesn’t just make you go faster, it also pitches the nose up, rolling doesn’t just roll, it also yaws in the opposite direction, and yawing rolls in the opposite direction, changing power can produce yaw as well, the stall speed is only valid under a specific set of conditions, and there is not normally any direct indication to the pilots of how close to the stalling angle they are. basically none of the indications you have are direct, and everything you do has a secondary effect. That’s not to say it’s difficult, because it’s not, but it is not very intuitive.

62

No but plummeting to the ground after the stall was till he recovered.

Did I mention he was a dick?

63

Yes, that is what I meant. The simulator companies do data gatherings for normal and limit-normal flight situations (say, including stall warnings, stick pusher, aerodynamic buffeting, single engine operation, etc) and the manufacturing companies will go further and test extreme cases with specialized aircraft and crew all equipped with parachutes (yes, even the plane!) and get data at the beginning of a stall, but they are all controlled stalls (barring the ones that failed, of course…sadly, there have been a few of those). That true airplane data can be - and is - combined with theoretical calculations to give the simulators as much data as possible to train pilots on, but there is no data for the uncontrolled events, like once AF447 really got in trouble, because past a certain point it’s just unsafe to do in real life. I suppose for some accidents the experimental/test plane for a given airframe could be equipped and tested in real flight situations to better understand an accident, but I’m not sure I’ve heard of any modern case where they have done so (I’m just an someone with an interest in aviation safety, not an investigator or anything).

The flight of AF447 will be tested in CatD sims and pilots/engineers will test responses to systems and conditions that keep the plane on the controlled - and data-supported - side of a stall, but it is all theoretical on the other side.

As a random aside, I had the opportunity to visit a sim once and the engineer was playing around and did some barrel rolls and loop-the-loops. He then “replayed” the flight from a third party perspective…and overlaid a B747 as the "airplane’ that performed the tricks, which was rather funny to see! The sims will allow you to do pretty much anything at all, if you turn of the real flight conditions. I then bounced the plane along the Thames, like a skipping stone. Rather amusing.

64

A lot of the information used in sims for the true edge of the envelope situations are taken from accident data, and black boxes are one of the sources. That’s one reason why black box recovery is still important, there can be data there that is completely unethical or impossible to obtain any other way.

Rather like studies have been done on peoples’ reactions to knowing they’re about to die based on cockpit voice recorders - an experiment and data otherwise completely impossible to obtain. (Trivia from that one: the most common last words of English-speaking pilots is “oh, shit.”)

This is the first real-world data ever obtained from an Airbus in a deep stall - a lot of people in aviation will want to see it, for a variety of reasons.

65

Question: Could an A-330 be recovered from a deep stall by full (or at least maximum safe) rudder and full power on the opposite side engine, with the goal of entering a spin?

(This presumes you could recover from a spin to normal flight in 30,000+ feet.)
ETA: For extra fun, perhaps you could use reverse thrust on the other engine.

66

I was curious about this, so I asked a friend who works for CAE about how accident data is used in full-flight CatD (or any category, reallly) simulators. In short (and as I understood him, any mistakes on this are my misinterpretation of a flurry of text messages), he says it isn’t used at all, barring the odd and unlikely request by particular clients to update existing simulators, something my friend doesn’t recall ever seeing or being a part of as part of his job.

The reason being that these things are fully qualified by the FAA/Transport Canada/Civil Aviation Authority of whatever country, and any change to the way the simulator functions would have to be re-qualified and re-certified by the regulatory body. These sorts of changes basically never happen because the time and cost to do so isn’t worth it, and because of the wonder of grandfather clauses: there are pilots out there being trained on 20 year old flight simulators that are not at the same standards as what’s coming out of the factory today (which isn’t to say the sims aren’t any good…as a comparison, note the 20 year old aircraft that are still deemed airworthy even though they aren’t (cannot be) manufactured according to today’s standards…it’s a matter of available technology at the time of certification). When a new simulator - say for an A330, for which previous sims exist, is made - the existing data model is basically copied and tuned to pass the initial qualification, it is not re-created brand new.

Black box data might be used to program a series of instructions/events in a simulator (night time, cross winds, loss of air speed indication, loss of #1 engine, whatever) in order to train pilots, but this is something done by instructors as operators of the machine, not done at the level of the machine’s software to control how the plane behaves under those circumstances (sorry…I feel I’m not being clear here and I’m having trouble wording this… it’s like the difference between me entering data into an Excel spreadsheet and Microsoft changing how Excel works, as another failed analogy on my part :))

67

Could someone elaborate on the reason why a deep stall is unrecoverable in one of these large aircraft (as has been stated in this thread)? I would naively think that given enough altitude and full throttle, the plane would slowly gain speed until controllable again. Is it that the AoA is too great for the thrust to overcome air resistance? And the elevator is useless during a stall?

My naive thinking is that during takeoff, engines at full thrust can bring the plane from zero to takeoff at high AoA in just 30 seconds, well short of the 3+ minutes the AF447 took to fall to the sea, so it seems like there should be plenty of thrust available to bring the airspeed up by brute force, regardless of the aerodynamic situation. What am I missing?

68

:eek:

In a spin it would come out of the sky even faster. I’d think a spin would just makes things orders of magnitude worse. A spin is just another type of stall, after all, and one where you drop much faster than most others.

I was thinking more along the lines of a research entity such as NASA using the information in that manner, or people developing new simulators, not retrofitting older ones. Sorry if that wasn’t clear.

Are you thinking you get out of a stall by application of power? If so, that’s where you’re mistaken. You get out of a stall by lowering the angle of attack. IF you have enough altitude, and you have a working flight control or two, you don’t need any power to get out of a stall. Gliders - that is, unpowered airplanes, stall and unstall with no problem, I know people who perform spins in giders with no issues. It really is all about the angle of attack.

If you don’t change the angle of attack all the engine power in the world won’t get you out of a stall. If you do change the angle of attack, then even if you have no engine you will exit the stall.

What power does, when properly applied, is reduce the amount of altitude lost during the full sequence of stall and recovery. This is, of course, important because smacking into the ground (or ocean) is Very Bad. But until the angle of attack changes the power isn’t helping.

The only time being stalled doesn’t matter is when you have sufficient thrust the aircraft is more rocket than anything else. Mike Melvill demonstrated that during his first X-prize flight when, during a vertical ascent, SpaceShipOne started spinning. He was still going up due to rocket thrust even though the wings were stalled, hence the spin (also an illustration that spins can be achieved at high speed and in any direction, including staight upwards). In that case, obviously, he got the thing unspun and unstalled and landed safely - an Airbus, however, is a different sort of thing and it doesn’t have the raw thrust to fly on engine power alone without regard to whether the wings are stalled or not.

69

You would need an enormous amount of thrust, it can be done but it’s the type of thing aerobatic aircraft with close to a 1:1 thrust to weight ratio can do. A big problem is that the thrust vector pitches the nose up which further increases the angle of attack, deepening the stall, if you don’t have elevator authority then you can’t counteract the thrust vector, if you do have elevator authority then you can recover without using thrust. Another problem is that the airflow to the engines in a deep stall is coming from an angle that it’s not really designed to take. There was a commuter jet that a couple of pilots were ferrying without passengers, they decided to see how high they could get it. They got up over 40,000 feet but it ran out of airspeed and stalled and the stall caused the engines to flame out and seize. They continued to cock things up by not telling ATC exactly what the problem was, by the time they owned up to the fact they had lost both engines, they didn’t quite have enough height to get to an airfield and they crashed and burned. They did manage to recover from the stall itself, I’m only using it as an example of how flying at the edge of the flight envelope can cause engine problems. At any rate, applying power is exactly the wrong thing to do to recover from a stall because of the pitching moment I mentioned above.

Edit: Powering out of stall can only really work with something that has a high power to weight ratio, and preferably a propeller that provides airflow over the elevator. Regardless, it is something only a very few highly powered aerobatic aircraft can do, and the odd military jet such as the Harrier.

Edit 2: What Broomstick said.

70

OK, so if you have control of the elevators, a deep stall is recoverable? Earlier in the thread I got the impression that a deep stall in a large aircraft is unrecoverable separate from any other factors. So what you are saying is that one of the biggest factors in the AF447 crash was the fly-by-wire system essentially failing catastrophically by preventing the pilots from using the elevators to recover from the stall?

71

No, by “elevator authority” I mean the elevator has enough airflow to be effective. In a deep stall the elevator can be blanketed by the turbulent air from the main wings, the pilots can move the elevator all they like but they don’t have enough airflow to be able to have any effect over the aircraft.

72

I realize that ideally the idea is to lower the AoA. But I got the impression earlier in the thread that it is impossible to lower the AoA in a large plane in a deep stall due to the stall making the control surfaces unusable. I may have gotten the wrong impression.

I guess I have a hard time reconciling this with the fact that in 30 seconds one of these planes can go from zero to 160 mph and take off at a high AoA. It seems (again, naively) that the plane has the raw thrust (in conditions in which the engines are intaking still air) to accelerate from zero mph (stall) to 160 mph (no stall at sea level) in less time than it takes the plane to fall from 35000 ft. I understand that the engines don’t operate efficiently at high AoA and at lower airspeed, but isn’t this more or less the same conditions as during takeoff?

73

OK, this was the assumption I made in my original question. But I guess you are saying that at high AoA the plane doesn’t generate enough thrust in the forward direction to gain enough airspeed such that the air becomes non-turbulent enough to regain elevator authority.

74

Not necessarily impossible, but you’re in unknown territory, you’ve become a test pilot at that stage and all you can do is try something and see if it works, if it doesn’t, try something else and keep trying until you run out of time. There’s a reason big jets have stick pushers, their stall characteristics are adverse enough that you basically never want to stall one, if you do then you’ve already lost the battle.

On the ground the engines are getting air straight into the intake and the airframe isn’t being asked to fly until it has lots of airspeed, so everything works fine. Don’t get confused between angle of attack and the pitch angle in the climb, although the aircraft is pitched up 15 degrees or more, the relative airflow hitting the wings and engines is only somewhere between 5 and 10 degrees, so it’s not a high angle of attack situation like a stall.

75

The answer is… it depends.

In order for the control surfaces to work there must be smooth and sufficient airflow over those surfaces. Turbulent airflow, such as in a stall, or insufficient airflow, such as with low airspeed, either sharply reduces the effectiveness of those controls or makes them useless.

So, saying a stall is “recoverable” means the controls are sufficiently effective to get the airplane back to normal flying. This doesn’t mean it happens instantly - the controls may be effective, but not as effective as they normally are. If you do not have sufficient altitude and/or time you may not be able to recover from a stall that is otherwise recoverable. (In flight training it’s beat into your head Thou Shalt Not Stall At Low Altitude for this reason) It is also possible to get into a situation where the the airflow over the control surfaces is either so turbulent, so lacking, or both that you will never be able to recover no matter how much altitude and time you have. If that happens you are doomed.

I am not qualified to determine which of those two possibilities applied in this crash. Regardless, they ran out of sky before they solved the problem.

I think you’re confusing “nose up” with “high angle of attack”. They’re not the same thing. The angle of attack is relative to the direction of flight and/or the angle at which the air hits the wing. This need not be parallel with the ground, and frequently isn’t. For this reason, you can stall on descent, or be non-stalled despite having a high angle between ground and nose.

In flight training they give you some empirical experience with this, as it is hard to understand for folks accustomed to traveling along the ground rather than through the air. Even then, it takes awhile to grasp this on a gut level.

What makes the air non-turbulent and restoring elevator authority is reducing the angle of attack, not any power coming from the engines.

Really, it’s all about the angle of attack.

If the angle of attack is high enough that you’re stalled applying more power isn’t going to help. Even if more air goes over the control surfaces it will still be too turbulent to allow those controls to be effective. Actually, applying more power when stalled usually just makes the problem worse - more violent turbulence and shaking, in some airplanes it can result in extreme rolls, flipping over, spins, faster spins…

I am not familiar enough with the Airbus line to know what the specifics effects of applying more power in a stall would be, but based on what I know of stalls I know that applying more power won’t solve the problem until the angle of attack is reduced.

76

Thanks you two for your replies. I’ve got a better feeling now for the problem.

77

True. But the goal should not be “Let’s see if we can delay our impact with the ocean for an additional minute or so” but rather “Let’s see if we can re-establish normal flight.”

In a deep stall, a direct and immediate recovery may not be possible. But if this can be converted to a maneuver (such as a spin) from which a recovery is possible, that would seem like a sensible thing to try.

And it’s at least possible that application of power may yield additional airspeed, additional airflow over the elevators, and thus improved ability to influence AoA.

It seems clear that in a deep stall with engines at low power, descending at 10,000 feet per minute, and nothing changing, the pilots should have been trying something different during those 3.5 minutes. They probably would have done so had they properly understood their situation.

78

So, an emergency rocket thruster above the cockpit, to shove the nose down?

Eh, no, it would probably punch right through the cockpit ceiling and kill the whole crew.

Dammit!

79

Before accepting your doom, you should probably consider that airflow over control surfaces could be affected by such things as:

  • use of engine thrust (including differential thrust)
  • deploying flaps and/or spoilers
  • lowering landing gear
80

On preview, I see that much of this has been covered by Broomstick and Richard Pearse, but I’ll post it anyways.

I think we need a clarification of terms here. Angle of Attack (AoA) and pitch attitude are two different things, as I believe Broomstick was attempting to explain with the X-Prize story. Imagine a line that passes from the nose through the tail of the airplane; we call this the longitudinal axis. Pitch attitude is the angle between the longitudinal axis and the horizon. So long as the horizon is clearly visible from inside the airplane, pitch attitude is fairly obvious to both pilot and passenger. From up front, we see less green and more blue out the window. From the back you notice your view of the world is rather askew. When we’re in the clouds or at night, we use the Attitude Indicator to depict the relationship between the airplane and the horizon. This instrument, as mentioned upthread, is front and center on the panel for each pilot, and at least triple redundant on large airplanes. In theory at least, pitch attitude has absolutely nothing to do with an aerodynamic stall.

Now imagine another line, this time drawn from the trailing edge of the wing straight forward through the leading edge. This is called the chord line. Angle of Attack is the angle between local airflow at the wing and the wing’s chord line. This angle, along with the velocity of airflow and shape of the wing, is what determines how much lift the wing makes. The higher the angle of attack, the more lift we can generate. Think about driving down the highway with your arm stuck straight out the window. If you flatten your hand and angle it up slightly from the horizontal, your arm will rise due to the lift generated by your hand. If you keep rotating your hand up, you will feel the lifting force increase. You are increasing the angle between your hand and the airflow, i.e. angle of attack, and generating more lift. The angle of attack is not apparent to passengers in an airplane (nor is it always apparent to the pilots either). Some airplanes do have AoA gauges, but typically pilots mainly use airspeed to judge angle of attack. Angle of attack has everything to do with an aerodynamic stall.

As a wing moves through the air, the flow over the top of the wing is accelerated aft and (slightly) downward. For a given velocity, the more we increase the angle of attack, the more dramatic this acceleration becomes. Eventually, we will get to such a high angle of attack that instead of smoothly flowing over the top of the wing in a nice curve, the airflow will separate from the top of the wing, rapidly and significantly reducing lift while at the same time increasing drag. This is an aerodynamic stall.

See NASA’s FoilSim. If you slowly increase the angle of attack (labeled Angle-deg), you will see that eventually the air no longer follows the top of the wing. The angle at which this happens is called the critical angle of attack, and that’s where that particular wing stalls.

Typically, in real life this happens because we let the airplane get too slow. The recovery is 100% all about reducing the AoA below the critical AoA, restoring smooth airflow to the top of the wing. We do this by using the elevators to lower the nose, which simultaneously reduces our angle of attack and takes advantage of gravity to help increase our airspeed (further reducing the AoA). In a deep stall, the rough airflow from the wings flows over the elevators, rendering them less effective. This may make it impossible to get the nose down for an effective stall recovery. Something like an F-15 fighter or a rocket powered airplane may have so much thrust available relative to mass that an application of full power might be sufficient to get you moving in the forward direction enough that airflow will be restored over the wings. However, an airliner is nowhere near capable of a stunt like that. Not only that, but as other have alluded to, an airplane like the A-330 has the engines mounted low under the wings. Because the thrust line is below the center of gravity, a power application will actually cause an increase in pitch (and angle of attack). As an added note, it is also possible to induce a stall at high speeds if we have enough pitch authority to rapidly raise the nose (quickly increasing AoA and lift) before our flight path has a chance to change. This is a good way to break the wings off.

A spin is when one wing is stalled more deeply than the other, which causes asymmetrical lift and drag. A spin is typically a Bad Thing in anything larger than a single engine trainer/aerobatic airplane. Recovery can range from simple to impossible in small airplanes, depending on configuration, design, loading, power setting, and pilot technique, among other things. A spin in an airliner would be absolutely unrecoverable and another great way to break the wings off.

To bring this all back to the accident in question, a sudden onset autopilot-induced high altitude deep stall at night over the middle of the ocean into a severe thunderstorm in a heavy swept-wing jet with a cockpit full of warning lights and horns and conflicting airspeed indications AND a simultaneous runaway elevator trim is about as bad a scenario as I can think of, to the point of being almost laughable. It’s certainly sobering to put myself into their seats and try to imagine how utterly disorienting, confusing, and frankly, terrifying, that it must have been.

StF